Can get more addictive ..Can it ?

[churchhaze]

Using the parts you used, I would have done it like this. (the one you said was probably the best).

However, The reason I strayed away from vero 29 in the first place is because I was designing for a lower tent/box profile. That is the reason I used 4 vero 18 to replace every 2 vero 29 for my newer design (the UFO killer). r-tools actually told me that putting the sources slightly closer together would have provided better temperature distribution, but I went a few milimeters more toward the perimeter. Still pretty close to perfect temperature distribution.

Now...
Please enlighten me ,with a better way ,placing two COBs in a small case ,
along with heat sink ,drivers ,fan psu and rest ...Give me your plan of better light distribution ,
with these parts .
Or better yet ,do not be restricted by parts .Show me any other or another alternative .
The one leaning towards 'better light power distribution. Please,give or share with me ,
even the slightest of your ideas .
Thank you.

Cheers.

 

[AquariusPanta]

Equates that something's wrong ...


Two fans ...
In series (one attached behind the other or in parallel ?
If in series |Flow rate remains same ,air pressure is doubled ..
If in parallel ,pressure remains same,flow rate doubles ...

I'm not sure if " flow velocity =rpm * radius of Fan " is correct ...
It does not seem to be ... ( It seems to be the speed of the rotor wings ,right at their tip )



The total area of the thermal transferring surface ..

For CXA3070 is the whole chip ( 27.35 x 27.35 mm ) ,
for the Vero29 it's thermal pad ( 31.8 x 31.8 mm )

Practice. Practice. & more practice for me!

My equation involves running 2x RoseWill 140mm fans, either 5V, 7V, or 12V. I guess that's a good place to start. If I ran the two fans in parallel (or should I reconsider series?) at 5V, which I'd initially prefer due to Supra's Law of Efficiency, how might that affect the stated CFM's and air flow?

Since RoseWill only offers air flow in CFMs and not m^3/h as you chose, Sailor, I'll pretend that the fans would produce the stated ~55 CFMs, ~1200RPMs while ran at the optimal 12V. Because I haven't worked with this sort of problem before, I'm assuming that by reducing the potential fed to the fan, from 12V to 5V, that I can expect the fans to run similarly to that of just one optimally powered (12V, etc.). By assuming, this would give me around 50CFMs, 1100RPMs that would be spread across my chosen HS. So for the flow rate section, I went ahead and entered 50CFM.

For clarity's sack, DigiKey lists the diameter of the LES to be a few millimeters smaller than what you listed. Not sure if this makes a huge difference in any given equation but I figured I'd bring it up.

Beginners Tip*: Use the drop-down option of generic when it comes to the thermal properties of each COB, like Sailor used. I goofed up early on and clicked one of the pre-selects and even with entering in my own numbers for PK3, it ended up giving my bizarre pressure drop numbers. It's the only variable that took me from ~80 Pa to ~20 Pa. I used thermal impedance instead of thermal conductivity also. I'll enter in the latter and see if it changes the thermal outcomes.

Edit: Ok, I tried playing around with the CFM. 110CFM gives me crazy high pressure drop (~100Pa), while 50CFM gives me 21Pa. Not really sure what to make of this. Going with the 110CFM lowers temperatures down to ~40C, while just 50CFM offers temperatures around 52C.

 

[churchhaze]

Just in case you happened to miss my design, this is the UFO killer design.

Sorry for the poor quality. It's an HLG-80H-C700A and 10"x10" sinks with 1 inch fin and 0.3in base.

When underdriven at 500mA, no fans are needed at all, otherwise, at 700mA, a light breeze will bring case temperature down from about 70C to 35C.

 

[stardustsailor]

Using the parts you used, I would have done it like this. (the one you said was probably the best).

However, The reason I strayed away from vero 29 in the first place is because I was designing for a lower tent/box profile. That is the reason I used 4 vero 18 to replace every 2 vero 29 for my newer design (the UFO killer). r-tools actually told me that putting the sources slightly closer together would have provided better temperature distribution, but I went a few milimeters more toward the perimeter. Still pretty close to perfect temperature distribution.

Now .I've used Vero 18 myself .
In upgrading an older system ,in order to be sold .
9 of them .( 3.5 K -80Ra )

My concerns / thoughts about that way :
Yes .Far greater way to use smaller power COBS (easier to find ,also ) ,in order to distribute light power better
than using less units of higher powered COBs. Another great plus is the easier cooling design of multiple of lower power units ,as you've already mentioned.

But ....

Driving ...

1st way :
Multiple AC/DC units ,one for each cob or more connected in series .
I do not like the 'in series' idea at all ,personally.
One AC/DC CC driver for each COB ,eats up space inside a case .Lots of it .
And adds further weight.

2nd way : external / remote AC/DC unit and in case one DC/DC CC buck per COB ...
OK ...Issues of !st way are solved.
But driving efficiency suffers. ( Total driving efficiency = AC/DC PSU eff. * DC/DC Buck eff.

Both ways : Lots of wiring .Lots .
Bad issue for designing / making encased electronics ..For my POV as Salm ,would say ..

But still ,yes ,that is indeed a better alternative .
With it's own issues .

Vero29 should be treated as if they were HID lights ...
They share plenty of similarities ..

Vero18 arrays are like a hybrid ..
Something between a high power COB and a single led unit ....

Yes ...While that way is offering both superb light distribution and cooling ,
electrical wiring & driving sucks big time ...

Nothing's perfect in this universe ..
Nor the universe itself ..


Light Alchemists Rule #557 :
" Perfect |" / Absolute can't co-exist with " choice / alternative " ,in this universe.


Cheers .

 

[AquariusPanta]

But ....

Driving ...

1st way :
Multiple AC/DC units ,one for each cob or more connected in series .
I do not like the 'in series' idea at all ,personally.
One AC/DC CC driver for each COB ,eats up space inside a case .Lots of it .
And adds further weight.

2nd way : external / remote AC/DC unit and in case one DC/DC CC buck per COB ...
OK ...Issues of !st way are solved.
But driving efficiency suffers. ( Total driving efficiency = AC/DC PSU eff. * DC/DC Buck eff.

Both ways : Lots of wiring .Lots .
Bad issue for designing / making encased electronics ..For my POV as Salm ,would say ..

But still ,yes ,that is indeed a better alternative .
With it's own issues .

Vero29 should be treated as if they were HID lights ...
They share plenty of similarities ..

Vero18 arrays are like a hybrid ..
Something between a high power COB and a single led unit ....

Yes ...While that way is offering both superb light distribution and cooling ,
electrical wiring & driving sucks big time ...

Nothing's perfect in this universe ..
Nor the universe itself ..


Light Alchemists Rule #557 :
" Perfect |" / Absolute can't co-exist with " choice / alternative " ,in this universe.


Cheers .

For my next big COB project, I was thinking of using a single, efficient, and expensive MeanWell driver to power multiple VERO 18's.

Tell me why you are disgruntled with the idea of running multiple COBs (each being of the same series/branch) in series?

 

[churchhaze]

The way I see it, as long as the cobs being run in series are all screw mounted to the same chunk of metal, series connections are safe. There is very little risk of the chain coming apart in the middle and shorting so long as each series chain is a unit.

The voltage in the holes in the wall here is around 170V peak. I consider about 200V and below to be relatively low voltage, and safe for household use. Obviously the Europeans use even higher and still don't hesitate to plug lamps directly into their shock ports.

The drivers I'm using (HLG-80H-C700A) go up to 129V output max. Not too scary, but it will kill you if you're not careful, like many other things.

Lightning bolts aren't going to start jumping off the uninsulated pads.

 

[stardustsailor]

For my next big COB project, I was thinking of using a single, efficient, and expensive MeanWell driver to power multiple VERO 18's.

Tell me why you are disgruntled with the idea of running multiple COBs (each being of the same series/branch) in series?

Not for any partricular reason.
I just do not like it .
Althought ,one can never know .
I might utilise it myself in the future.

It's a very good way.
Stay on the plan ,you've made .
Looks more than promising .

Cheers.

 

[AquariusPanta]

Something I stumbled upon while looking for answers about pressure and flow.

http://www.thermalsoftware.com/optimum_sink_fan.pdf

 

[Voidling]

I wish I could take a class from you on making lights like you do. Very nice work

 

[SupraSPL]

Just measured the LES of the Vero 29 to double check for future reference, actual LES is 29 millimeters.

Regarding spacing on the heatsink. I was thinking along the same lines as churchhaze to optimize spread wherever possible. I am repurposing heatsinks that are not giving me the optimal spread, but it is close enough for now.

Originally spaced evenly, about 5"-6" spread on a 10" heatsink.


Staggered diagonally, gave me 7" of spread. Maybe a small sacrifice thermally, but if you have a thick baseplate and a low thermal load, should be no worries.


Anyway, neither option is a good as using a heatsink that gives you the correct spread in the first place, but you have to work with what you've got. SDS, are reflectors a consideration for this design?

 

[stardustsailor]

Just measured the LES of the Vero 29 to double check for future reference, actual LES is 29 millimeters.

Regarding spacing on the heatsink. I was thinking along the same lines as churchhaze to optimize spread wherever possible. I am repurposing heatsinks that are not giving me the optimal spread, but it is close enough for now.

Originally spaced evenly, about 5"-6" spread on a 10" heatsink.
View attachment 3312852

Staggered diagonally, gave me 7" of spread. Maybe a small sacrifice thermally, but if you have a thick baseplate and a low thermal load, should be no worries.
View attachment 3312854

Anyway, neither option is a good as using a heatsink that gives you the correct spread in the first place, but you have to work with what you've got. SDS, are reflectors a consideration for this design?

LES of Vero 29 : ~29.2 mm Dia.


Reflectors can be placed .
Depends mainly on the available space ,it's actual 'shape' and it's surroundings reflectivity.
Another factor determining the use of optics is the cultivated technique(s) utilised .
( tree-style ,SOG/SCROG, bed ,vertical and/or other ).

Cheers.

.

 

[AquariusPanta]

View attachment 3312901

LES of Vero 29 : ~29.2 mm Dia.


Reflectors can be placed .
Depends mainly on the available space ,it's actual 'shape' and it's surroundings reflectivity.
Another factor determining the use of optics is the cultivated technique(s) utilised .
( tree-style ,SOG/SCROG, bed ,vertical and/or other ).

Cheers.

.

Just measured the LES of the Vero 29 to double check for future reference, actual LES is 29 millimeters.

For clarity's sack, DigiKey lists the diameter of the LES to be a few millimeters smaller than what you listed. Not sure if this makes a huge difference in any given equation but I figured I'd bring it up.

This is why teams work best, because everyone (no matter how smart or experienced) makes errors from time to time. It must be apart of our nature . By the way, the VERO 18's LES is 18mm, hence VERO *18* .

Anyhow, I'm under the impression that the higher the pressure drop (change in pressure), the harder it is for fans to circulate air flow through the inner fins of the HS - as they can't keep up due to the resistance of the HS. Mathematically, a designer of an actively cooled HS would want 17Pa (Pascals for you non-engineers) of pressure drop rather than 80Pa. The thing though is that with 50CFM, I get former pressure drop, while if I double my CFM, I end up with the latter pressure drop. I suppose it's just too much if I run two 140mm fans at full throttle (12V)?

If I got this backwards, please correct me.

About the reflectors - they seem sort of hard to come by (for VERO 18 ) and when found, cost quite a little. While I'd opt for ~90degrees over a 120degrees spread, the costs are still too high. I'll continue searching for an affordable and reliable source for the VERO line in terms of reflectors in the meantime.

 

[SupraSPL]

I have been making them out of soda bottles and taping them on with kapton. Just for experimental purposes but it gets the job done. Clearly could use some more paint on this one.
This is a 5000K CXA3590 vegging some 4 foot tall ChemD X GSC. It is dissipating about 18W @ 62% efficient. For the record I am opposed to soda

 

[AquariusPanta]

I have been making them out of soda bottles and taping them on with kapton. Just for experimental purposes but it gets the job done. Clearly could use some more paint on this one.
This is a CXA3590 vegging some 4 foot tall ChemD X GSC. It is dissipating about 18W.
View attachment 3312955

You and that Kapton Supra LL. I gotta find me a roll of that stuff!

I take it your painting the insides (and outsides?) with white paint, otherwise the light would most likely escape through the plastic?

 

[SupraSPL]

Yep spray painting the inside with white. The reflector in that pic only got a thin coat. I am using Rustoleum professional semi-gloss white 239108 (titanium dioxide of course)

I have been wondering about how exactly the reflection works. Whichever photons penetrate the upper layer of paint will be scattered inside. So I am wondering if extra layers will even make a difference. In other words I suspect that the only photons we can recover need to bounce off the surface and get lucky enough to bounce in the direction of the canopy. The vast majority of the ones that do not bounce off the upper layer are probably lost no matter how thick the layer of paint?

 

[AquariusPanta]

Yep spray painting the inside with white. The reflector in that pic only got a thin coat. I am using Rustoleum professional semi-gloss white 239108 (titanium dioxide of course)

I have been wondering about how exactly the reflection works. Whichever photons penetrate the upper layer of paint will be scattered inside. So I am wondering if extra layers will even make a difference. In other words I suspect that the only photons we can recover need to bounce off the surface and get lucky enough to bounce in the direction of the canopy. The vast majority of the ones that do not bounce off the upper layer are probably lost no matter how thick the layer of paint?

The upper layer of paint? Hmm. I would think that as long as the paint coats the plastic, the photons will have no choice but to be reflected downward below. It's a good point to bring up, because I'm currently unaware of how thick a reflective material has to be before photons start bouncing off the material.

Could one expect health hazards with using titanium dioxide with heat from the HS/COB being a factor?

Also, would the titanium dioxide absorb any UV rays emitting from the COB and is that a concern for you? (I'm assuming not)

 

[PSUAGRO.]

Yep spray painting the inside with white. The reflector in that pic only got a thin coat. I am using Rustoleum professional semi-gloss white 239108 (titanium dioxide of course)

I have been wondering about how exactly the reflection works. Whichever photons penetrate the upper layer of paint will be scattered inside. So I am wondering if extra layers will even make a difference. In other words I suspect that the only photons we can recover need to bounce off the surface and get lucky enough to bounce in the direction of the canopy. The vast majority of the ones that do not bounce off the upper layer are probably lost no matter how thick the layer of paint?

Smart man^^^..............but why not go with flat white supra????

 

[churchhaze]

Nice idea. I was trying to do something similar with the bottom of an aluminum can, but those aren't even shaped well!

Why not silver paint? (totally 180 degree question from PSUAGRO)

 

[SupraSPL]

Based on how bright the reflector is from the outside, it seems quite a lot of the photons fail to bounce off the upper layer, but penetrate and end up trapped inside the paint/plastic and many pass through entirely.

Good suggestions. I have been thinking we need a "reflectors" thread and we can do some testing and to if any of these options stand out above the others.
Aluminum
Flat white
Satin white
Semi gloss white
Gloss white
Silver/metallic paint
Thin layer vs thick layer
etc

The reason I tried semi gloss first was based on the idea that photons hitting the reflector at a high angle of incidence would skim right down into the canopy rather than being scattered, but photons with a low angle will experience diffuse scattering and many will not make it to the canopy.

 

[stardustsailor]

Regarding thermal simulation and calculations of the fixture ...

Paying a tad more attention to the Vero29 new updated datasheet ,
the thermal resistance of junction -case is not 0.28 °C/W ,but 0.13 C/W .


making a new sim with the new Rth j-c of 0.13 ..



The results ..
...


T heat sink= 52.5 °C
T case = 54 °C
T junction = 61.4 °C ....


When Ta = 25 ° C ,Noctua fan is working at it's full spec and the two veros @9 are driven at 2100 mA .
According to this Simulation software ...

Let me confirm that with some math ...

I need some thermal resistances figures ..
- Rth heatsink-ambient
- Rth case- heatsink ( aka Rth TIM )
& and the one known
-Rth junction -case = 0.13 °C/W


Rth heatsink-ambient

Well ..
There are some other brand models that are exactly identical to the heatsink utilised..

The heatsink used is a " ST20 model made from a local brand .

At 200 mm lenght it is stated to have a thermal resistance of 0.3 °C/W ,
when passive cooling .

Another almost identical heatsink like the one used,
is from ABL heatsinks ...
Model 159AB

http://www.abl-heatsinks.co.uk/index.php?page=extrudedproduct&product=13

Also stated at 200 mm length ,having a thermal resistance figure of 0.36 °C/W when passive .
The ABL is offering a th. resistance calculator ...
But in order to find the th. resistance of the heatsink when active cooling,
the air velocity must be known and set ...

Now ..The Noctua fan used is having a flow of ~110 m^3/ h ..
And a rotor diameter of 130 mm .
Still,it is placed on top of the heatsink ,fin-side and at center .
The heat sink is forming two exausts with the top lid ...
So the air flow coming from top of fin side ,it 'divides' towards the two exausts / sides of the heatsink ...
at each side ,it is assumed that the flow is half of the one supplied by the fan ...
That is the airflow through the fins ,from top & center of heat sink towards the sides ,exausting .

So circular duct of R=65 mm (fan ) and airflow of 55 m^3/h .

Set at this calc tool :
http://www.engineering.com/calculators/airflow.htm

To get a result of air velocity = 1.15 m/sec

Onto the ABL's calc tool ...



So Rth heatsink -ambient ,when active cooled ,is ~0.25 °C/W .


- Rth case- heatsink ( aka Rth TIM )

Arctic Silver 5 is used ,having a thermal conductivity of 8.7 W/m.K and at a thickness of 150 microns approx.

Area of thermal transfer area is 31.8 mm x 31.8 mm ...

Thermal conductivity = ( Heat power * TIM thickness ) / ( heat transfer area * ΔΤ )
8.7 = 1 * 0.00015 / 0,00101124 * ΔΤ =>
ΔΤ= 0.017
Rth =ΔΤ / heat power = 0.017 / 1 = 0.017 °C/ W

Now I have all three thermal resistances needed ...

- Rth heatsink-ambient =0.25 °C/W
- Rth case- heatsink ( aka Rth TIM ) = 0.017 °C/W
- Rth junction - case = 0.13 °C/W

T ambient =25 C
T heat sink = 25 + ( 0.25 * ( 2 *50 Watt ) )= 50 C
T case = 50 + ( 0.017 *50 ) = 50.85 C
T junction = 50.85 + (0.13 *50 )= 57.35 C

Small difference between results is because of slightly (?) different heatsinks used in the 3D sim and in math calculation ...

Giving some ' tolerance ' (because of TIM application ,pressure drop of fan ,etc ) say +5 C ,
the cooling set up of the fixture ,will keep T case of both Veros driven at 2100 mA ,
around 55 C ,when T ambient is 25 C . Both having a T juncton around 55 + 6.5 = 61.5 C

And now ...
How useful is that 55 C ,T case figure ....
Back to Veros Datasheet,then ...


Cheers